Maybe it would help to apply the temperature first (with creep effects turned off) and then turn on creep in a subsequent load step and use smaller time increments. With creep off, can you verify how much creep strain you expect for a given stress state and time increment (by hand calc). Maybe that would he insightful.